CN114279973B - In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network - Google Patents
In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network Download PDFInfo
- Publication number
- CN114279973B CN114279973B CN202111610089.XA CN202111610089A CN114279973B CN 114279973 B CN114279973 B CN 114279973B CN 202111610089 A CN202111610089 A CN 202111610089A CN 114279973 B CN114279973 B CN 114279973B
- Authority
- CN
- China
- Prior art keywords
- neural network
- artificial neural
- temperature
- variable
- soil body
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000002689 soil Substances 0.000 title claims abstract description 136
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 87
- 230000001052 transient effect Effects 0.000 title claims abstract description 58
- 239000000835 fiber Substances 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000012544 monitoring process Methods 0.000 title claims abstract description 36
- 238000011065 in-situ storage Methods 0.000 title claims abstract description 26
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 61
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 238000012546 transfer Methods 0.000 claims abstract description 11
- 238000010606 normalization Methods 0.000 claims description 26
- 230000008859 change Effects 0.000 claims description 17
- 210000000476 body water Anatomy 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 7
- 238000013434 data augmentation Methods 0.000 claims description 3
- 238000001035 drying Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 2
- 238000013508 migration Methods 0.000 claims description 2
- 230000005012 migration Effects 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000003062 neural network model Methods 0.000 abstract description 3
- 238000010438 heat treatment Methods 0.000 description 20
- 230000000694 effects Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000003672 processing method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses an in-situ monitoring method for the moisture content of a soil body of a transient variable temperature fiber grating based on an artificial neural network. An artificial neural network model is established, an input layer of the model is set as a variable-temperature characteristic index of a measuring point and an adjacent measuring point, and an output layer is the water content of a soil body of the measuring point. Training the constructed artificial neural network model by using a training data set, testing the trained model by using a testing data set, and taking the neural network model meeting the expected error as a calibration model. And inputting the measured variable-temperature characteristic index into a calibration model to achieve the aim of in-situ monitoring of the water content of the soil body. The invention reduces the error caused by the longitudinal heat transfer inside the transient temperature-changing fiber grating sensor, and the obtained calibration model has strong applicability and high accuracy of the monitored soil moisture content.
Description
Technical Field
The invention relates to a soil moisture content monitoring method, in particular to a soil moisture content in-situ monitoring method of a transient variable-temperature fiber bragg grating based on an artificial neural network.
Background
In recent years, the transient temperature-changing fiber bragg grating sensor is widely applied to monitoring of the water content in soil, mainly comprises a high heat conduction material, a resistance wire, fiber bragg gratings FBG and the like, and has the advantages of simplicity and convenience in installation and operation, small disturbance to the soil, electromagnetic interference resistance, corrosion resistance, good stability and the like.
In addition, the transient temperature-changing fiber bragg grating sensor can be connected with a plurality of FBGs in series according to the requirements of monitoring density and measuring points, so that the quasi-distributed monitoring of the water content of the soil body is realized. The calibration work needs to be completed before the transient temperature-changing fiber grating sensor performs quasi-distributed monitoring, and if the water content distribution gradient of the soil around the transient temperature-changing fiber grating sensor is smaller during calibration, the temperature changing process inside the transient temperature-changing fiber grating sensor can be approximately considered to be basically consistent, namely the heat transfer phenomenon along the pipe body of the transient temperature-changing fiber grating sensor does not exist inside the soil. However, in practice, the moisture content of the soil body at the adjacent measuring points often has a large difference, namely different heat conductivity coefficients, so that the temperature change inside the transient variable temperature fiber bragg grating sensor has a difference, thereby causing the longitudinal heat transfer inside the transient variable temperature fiber bragg grating sensor, and enabling the variable temperature characteristic index of the FBG to be influenced by the properties of the adjacent soil body.
At present, the traditional calibration method of the transient temperature-changing fiber bragg grating sensor is low in calibration speed, low in precision and poor in applicability, only a traditional empirical formula method is adopted for fitting and calibrating the temperature-changing characteristic index at one measuring point and the soil moisture content value, and the influence of the longitudinal heat transfer effect inside the transient temperature-changing fiber bragg grating sensor is not considered, so that the error of a calibration model is increased, and the precision of monitoring the soil moisture content is influenced.
Therefore, how to improve the accuracy of monitoring the water content of the soil body by the transient temperature-changing fiber bragg grating sensor through a calibration method becomes a technical problem to be solved urgently.
Disclosure of Invention
The invention aims to: the invention provides an in-situ monitoring method for the moisture content of a soil body based on a transient variable temperature fiber grating of an artificial neural network.
The technical scheme is as follows: the invention discloses an in-situ monitoring method for the water content of a soil body based on a transient variable-temperature fiber bragg grating of an artificial neural network, which comprises the following steps:
(1) Obtaining calibration data, namely obtaining a variable temperature characteristic index at a measuring point through a transient variable temperature fiber bragg grating sensor, and obtaining a true water content value of a soil body at the measuring point through a drying method or a water content sensor;
(2) The method comprises the steps of constructing and dividing a data set, namely taking a variable-temperature characteristic index at a measuring point and a true water content value of a soil body as the data set, and dividing the data set into a training data set and a testing data set;
(3) Carrying out data set normalization processing, specifically, carrying out normalization processing on the variable-temperature characteristic index of the training data set, the true value of the water content of the soil body and the variable-temperature characteristic index of the test data set;
(4) Establishing an artificial neural network model, training, establishing a calibration model suitable for the transient variable temperature fiber bragg grating sensor based on the artificial neural network model, and training a normalized training data set to obtain a trained artificial neural network model;
(5) The artificial neural network prediction is carried out, specifically, normalized variable-temperature characteristic indexes of the test data set are input into a trained artificial neural network model, and a soil body water content normalized value predicted by the artificial neural network is obtained;
(6) Performing return normalization and error calculation on the soil moisture content value, namely performing return normalization treatment on the soil moisture content normalization value predicted by the artificial neural network, and calculating Root Mean Square Error (RMSE) between the return normalized soil moisture content value and the soil moisture content true value;
(7) Performing artificial neural network model verification, setting expected RMSE, and repeating the steps (4) to (6) if the RMSE in the test step (6) is greater than the expected RMSE; if the RMSE in the test step (6) is smaller than or equal to the expected RMSE, taking the trained artificial neural network model as a calibration model of the measuring point;
(8) Determining a calibration model, namely, completing calibration of all measuring points to obtain a soil moisture content calibration model of the transient variable-temperature fiber bragg grating sensor based on the artificial neural network;
(9) And inputting the measured variable-temperature characteristic index into a soil moisture content calibration model, and carrying out in-situ monitoring on the soil moisture content.
The temperature change characteristic index in the step (1) is obtained by processing a temperature change value measured by a transient temperature change fiber grating soil body water content sensor, and the temperature change characteristic index is divided into a temperature increase characteristic index and a temperature decrease characteristic index, wherein the temperature increase characteristic index processing method comprises a maximum temperature increase value method, a temperature characteristic value method and an accumulated temperature increase value method, and the temperature decrease characteristic index processing method comprises a logarithmic slope method and a temperature decrease rate method.
The training data set in the step (2) is used as a data set for training an artificial neural network; and in the test data set, carrying out error analysis on the predicted value of the water content of the soil body obtained by the trained artificial neural network model.
Normalizing the data set in the step (3) to normalize the variable-temperature characteristic index and the true value of the water content of the soil body to be numerical values between [0,1], so as to accelerate the training rate of the constructed artificial neural network model and improve the convergence of the artificial neural network model, wherein the specific expression is as follows:
wherein: y is ΔT (i) Is the temperature change characteristic index after the i group normalization, y θ (i) The water content value of the soil body after the normalization of the ith group is obtained; delta T i For the i-th group of variable-temperature characteristic indexes in the data set, delta T max For the i-th group maximum variable temperature characteristic index in the data set, delta T min The method comprises the steps of (1) setting the minimum variable-temperature characteristic index of an ith group in a data set; θ i For the water content of the soil body of the ith group in the data set, theta max For the maximum soil moisture content in the data set, theta min Is the minimum soil moisture content in the data set.
In the step (4), the established artificial neural network model is divided into an input layer, an hidden layer and an output layer, the input layer of the artificial neural network model is set as a variable-temperature characteristic index of a measuring point and an adjacent measuring point, and the output layer is the water content of a soil body of the measuring point; training the modeling type by using a training data set, testing the training model by using a test data set to obtain Root Mean Square Error (RMSE) of a soil body water content predicted value, and taking a neural network model with the error smaller than or equal to an expected RMSE as a calibration model of a measuring point. By the method, measurement errors caused by heat transfer inside the sensor are reduced, a high-precision soil moisture content calibration model of the transient variable-temperature fiber grating sensor is obtained, and applicability is enhanced.
The purpose of inputting the variable-temperature characteristic index of the adjacent measuring points is as follows: and a transmission rule of the longitudinal temperature inside the transient variable-temperature fiber bragg grating sensor is established, the influence of the longitudinal temperature transmission on the variable-temperature characteristic index of the measuring point is removed, and the monitoring precision of the water content of the soil body is improved.
In the step (4), the established artificial neural network model is optimized by adjusting the number of nodes of the hidden layer and the learning rate.
In the step (4), the training process of the artificial neural network model is divided into forward and backward propagation; in forward propagation, the input layer is processed by the unit nodes of the hidden layer and is transmitted to the output layer, and if the output layer cannot obtain the expected output, the error signal is reversely propagated and returned along the original unit nodes of the hidden layer.
In the step (4), when the artificial neural network model is trained, a mapping relation between a variable-temperature characteristic index and a water content true value is established by using a transfer function, wherein the transfer function is a linear or nonlinear transfer function, and the expression is as follows:
wherein y is ΔT(s) Laplacian transformation, y, for normalizing temperature change characteristic index θ(s) Laplacian transform, normalized to the water cut value.
Adopting a gradient descent method to carry out iterative solution so as to minimize a loss function, thereby optimizing an artificial neural network model, wherein the expression of the loss function is as follows
L (θ,θ') ＝w (θ) (y θ -y' θ ) 2 (4)
Wherein: w (w) (θ) Is the weight; y is θ Normalizing the soil moisture content value; y' θ And outputting the output value of the artificial neural network model.
The soil body water content value normalization processing in the step (6) is to restore the soil body water content value normalization value to the original order of magnitude, and the expression is as follows:
θ 0 '(i)＝y θ '(i)×(θ max -θ min )+θ min (5)
wherein: θ 0 ' (i) is the return normalization value of the water content of the soil body of the i group; y is θ And (i) is an ith group of soil body water content normalized value obtained based on the artificial neural network model.
In the step (6), the expression of the Root Mean Square Error (RMSE) between the return normalized soil moisture content value and the soil moisture content true value is as follows:
wherein: m is the total number of groups of the test dataset, θ 0 (i) And (5) the true value of the water content of the soil body in the ith group.
In the step (7), when the steps (4) to (6) are repeated, the node weight and the threshold between the input layer and the output layer are readjusted when the artificial neural network model is trained.
The temperature change characteristic index in the step (9) needs to be the same as the transient temperature change acquisition mode and the processing method adopted by the temperature change characteristic index in the step (1).
The accuracy of predicting the water content of the soil body by the artificial neural network model is improved by increasing the number of training data sets. If the number of training data sets is rare, the accuracy of predicting the water content of the soil body by the artificial neural network model is improved through a data augmentation, model regularization and migration learning method. The data augmentation method is realized by increasing the number of training data sets, model regularization prevents the training data sets from being too few to cause overfitting, and the transfer learning establishes a new model by adjusting weights of the trained artificial neural model.
Working principle: the artificial neural network method has excellent self-organization, self-learning and self-adaption capability, and can make up for the defects of the transient variable temperature fiber bragg grating sensor in the calibration process. According to the invention, an artificial neural network method is adopted to establish a calibration model of the variable temperature characteristic indexes of the measuring points and the adjacent measuring points and the moisture content of the soil, and the internal connection of the heat transfer inside the sensor is explored by training the variable temperature characteristic indexes of the measuring points and the adjacent measuring points and the moisture content of the soil of the measuring points, so that the calibration precision of the transient variable temperature fiber grating sensor is improved, and the method has important significance for high-precision in-situ monitoring and research of the moisture content of the soil.
When the transient variable temperature fiber grating sensor is adopted for monitoring, the sensor embedded in the soil to be monitored is required to heat and cool with transient constant power, the soil with different soil moisture contents has different variable temperature response characteristics due to the difference of heat conductivity coefficients, the temperature of the soil can be monitored through the Fiber Bragg Gratings (FBGs) to obtain different variable temperature characteristic indexes, a calibration model of the variable temperature characteristic indexes and the true value of the soil moisture contents is established, and the variable temperature characteristic indexes are input into the calibration model, so that the effect of in-situ monitoring of the soil moisture contents is achieved, and the method comprises the following specific procedures:
the temperature change characteristic index of the measuring point is obtained through a transient temperature change fiber bragg grating sensor, and the true value of the water content of the soil body of the measuring point is obtained through a drying method or a water content sensor. And taking the variable-temperature characteristic index of the measuring point and the true value of the water content of the soil body as data sets, and dividing the data sets into training data sets and test data sets. And establishing an artificial neural network model, setting an input layer of the model as a variable-temperature characteristic index of a measuring point and an adjacent measuring point, and setting an output layer as the water content of a soil body of the measuring point. Training the constructed artificial neural network model by using a training data set, testing the trained model by using a testing data set to obtain Root Mean Square Error (RMSE) of the soil body water content predicted value, and taking the neural network model with the error meeting the expected RMSE as a calibration model of the measuring point. And inputting the measured variable-temperature characteristic index into a calibration model to achieve the effect of in-situ monitoring of the water content of the soil body. By the method, measurement errors caused by longitudinal heat transfer inside the sensor are reduced, the obtained calibration model is high in applicability, and the monitored soil body water content is high in accuracy.
The beneficial effects are that: compared with the existing transient temperature-changing fiber grating sensor calibration method, the method has the following advantages:
(1) The invention has the advantages of high calibration speed, high precision and simple operation.
(2) According to the invention, the input layer of the artificial neural network model is set as the variable temperature characteristic index of the measuring point and the adjacent measuring point, the transmission rule of the longitudinal temperature inside the transient variable temperature fiber bragg grating sensor is established by inputting the variable temperature characteristic index of the adjacent measuring point, the influence of the longitudinal temperature transmission on the variable temperature characteristic index of the measuring point is removed, the measurement error caused by the longitudinal heat transmission inside the transient variable temperature fiber bragg grating sensor is reduced, and the accuracy of in-situ monitoring of the water content of the soil body is improved.
(3) The invention expands the calibration range of the transient temperature-changing fiber grating sensor and improves the applicability of the calibration method.
Drawings
FIG. 1 is a flow chart of an in-situ monitoring method for the water content of a soil body based on a transient variable-temperature fiber bragg grating of an artificial neural network;
FIG. 2 is a schematic diagram of an artificial neural network model employed in the present invention;
FIG. 3 (3 a) is a diagram of a calibration test apparatus in the example; fig. 3 (3 b) is a temperature change characteristic index and soil moisture content truth value distribution cloud chart in the embodiment;
fig. 4 (4 a) is a root mean square error statistical chart of the embodiment based on the artificial neural network monitoring the moisture content of the soil body; fig. 4 (4 b) is a root mean square error statistical chart of the water content of the monitored soil body by the conventional calibration method.
Detailed Description
As shown in fig. 1-3, the data set identified in this example was from an in situ test. The transient temperature-changing fiber grating sensor selected in the embodiment has 5 FBG temperature measuring points, as shown in (3 a) of fig. 3, the distance between the temperature measuring points is 10cm, and the temperature measuring points are respectively buried in the soil body at the depths of 5, 15, 25, 35 and 45 cm. And placing an FDR (frequency domain reflectometry, frequency domain reflection) water content sensor at the corresponding depth of the temperature measuring point for monitoring the real soil water content of the soil. By controlling the distribution of the moisture content of the soil body, the transient temperature change is performed on the transient temperature change fiber grating sensor, the heating time is 12min, the power is 40W/m, the power and the time of each heating are the same, and therefore the temperature values corresponding to the moisture content of the soil body at different depths are obtained, and the result is shown in (3 b) of fig. 3.
In this embodiment, the temperature value measured by the transient variable temperature fiber grating sensor is processed by adopting a maximum heating value method in the variable temperature characteristic index, and the maximum heating value is used as the variable temperature characteristic index, and the formula is as follows:
ΔT＝T t -T 0 (1)
wherein DeltaT is the maximum heating value; t (T) t Is the temperature after transient temperature change; t (T) 0 Is ambient temperature.
In the embodiment, 40 groups of maximum heating value and soil moisture content true value are obtained, and the maximum heating value and soil moisture content true value are used as a data set. And randomly selecting 20 groups of data sets from the maximum heating value and soil water content truth value data set as training data sets, and using the rest 20 groups of data sets as test data sets.
Carrying out normalization processing on the maximum heating value of the training data set, the true value of the water content of the soil body and the maximum heating value of the test data set to obtain an expression:
wherein: y is ΔT (i) For the maximum heating value after the i group normalization, y θ (i) The water content value of the soil body after the normalization of the ith group is obtained; delta T i For the maximum heating value of the ith group in the training data set, θ i And (5) the true value of the water content of the soil body of the i group in the training data set is obtained.
An artificial neural network model is established based on a MATLAB neural network tool box and is divided into an input layer, an hidden layer and an output layer, wherein the input layer of the artificial neural network model is a variable-temperature characteristic index of a measuring point and an adjacent measuring point, and the output layer is a true value of the water content of a soil body of the measuring point.
Specifically, as shown in table 1, in this embodiment, the input layer of the model is the maximum heating value of 5cm and 15cm measuring points, the output layer is the true value of the water content of the soil body of 5cm measuring points, and the artificial neural network training is performed to obtain the trained artificial neural network model.
In fig. 2, the trained artificial neural network model is provided with two parts of forward propagation of input and backward propagation of error, a mapping relation between an input layer and an output layer is established, and the trained artificial neural network model completes the relation training of the maximum heating value of a measuring point and adjacent measuring points and the true value of the soil water content of the measuring point.
Table 1: input layer and output layer information table for calibrating measuring points
Then carrying out return normalization processing on the soil body water content normalization value predicted by the artificial neural network, wherein the expression is as follows:
θ 0 '(i)＝y θ '(i)×(0.418-0.003)+0.003 (4)
wherein: θ 0 ' (i) is the return normalization value of the water content of the soil body of the i group; y is θ And (i) is an ith group of soil body water content normalized value obtained based on the artificial neural network model.
The Root Mean Square Error (RMSE) between the return normalized soil moisture content value and the soil moisture content true value is calculated, and the expression is as follows:
setting the desired RMSE' to 0.01m 3 ·m -3 If RMSE > RMSE', reestablishing the artificial neural network model and training, and readjusting the node weight and the threshold between the input layer and the output layer when the artificial neural network model is trained. After multiple times of building and training of the artificial neural network model, the RMSE=0.009 m at the 5cm measuring point is obtained 3 ·m -3 And (3) taking the trained artificial neural network model as a calibration model of the 5cm measuring point, wherein the artificial neural network model is smaller than RMSE'. Under the condition that the RMSE is less than or equal to the RMSE' requirement, the trained artificial neural network model can be a calibration model.
And respectively establishing an artificial neural network for 15cm, 25cm, 35cm and 45cm measuring points and training to obtain a neural network calibration model of the corresponding measuring point, thereby completing the calibration model of the transient variable-temperature fiber bragg grating sensor. And inputting the maximum heating value measured by the in-situ transient heating fiber grating sensor into a calibration model to obtain the real soil moisture content, thereby achieving the purpose of high-precision in-situ monitoring of the soil moisture content. The root mean square error of the water content of the soil body of each measuring point of the model after training is shown in table 2.
Table 2: RMSE expected and obtained at different measuring points
And (3) re-dividing 40 groups of maximum heating value and soil water content true value data sets according to the table 3, and respectively carrying out the soil water content calibration method on the selected groups 1, 3 and 4 to obtain a soil water content calibration model of each group of transient variable-temperature fiber grating sensors, wherein the result is shown in fig. 4 (a).
Table 3: training and testing data set group table
The more the total group number of the training data sets is, the number of the training data sets is increased to reduce the root mean square error of the water content of the soil body predicted by the artificial neural network. Therefore, if a soil moisture content calibration model of the transient temperature-changing fiber bragg grating sensor with higher precision is obtained, a plurality of temperature-changing characteristic indexes and soil moisture content data sets as much as possible need to be obtained as training data sets.
The transient variable temperature fiber grating sensor is calibrated by adopting a traditional calibration method, namely, a calibration formula is used:
ΔT＝a·θ b (6)
and (3) fitting the maximum heating value delta T of the training data set and the soil water content true value theta by adopting a formula (6) to obtain the values of the parameters a and b, and establishing an empirical relationship between the maximum heating value and the soil water content to obtain the traditional calibration model.
Inputting the maximum heating value of the test data set into the calibration formula (6) to obtain a predicted soil moisture content value, and calculating the Root Mean Square Error (RMSE) between the predicted soil moisture content value and the soil moisture content true value of the test data set, wherein the result is shown in a figure (4 b).
Comparing the graphs (4 a) and (4 b), the RMSE range of the artificial neural network monitoring method is 0.002-0.011 m 3 /m -3 The accuracy of measuring the water content of the soil reaches +/-0.02 m 3 /m -3 Whereas the conventional calibration monitoring has an RMSE range of 0.017-0.048 m 3 /m -3 The precision is only + -0.08 m 3 /m -3 . Therefore, the method based on the artificial neural network meets the requirement of high-precision in-situ monitoring of the water content of the soil body of the transient variable-temperature fiber bragg grating sensor.
Claims (8)
1. An in-situ monitoring method for the moisture content of soil based on a transient variable-temperature fiber bragg grating of an artificial neural network is characterized by comprising the following steps: the method comprises the following steps:
(1) Acquiring a variable temperature characteristic index of a measuring point through a transient variable temperature fiber bragg grating sensor, and acquiring a true value of the moisture content of a soil body of the measuring point through a drying method or a moisture content sensor;
(2) Taking the variable-temperature characteristic index of the measuring point and the true value of the water content of the soil body as data sets, and dividing the data sets into training data sets and test data sets;
(3) Normalizing the variable-temperature characteristic index of the training data set, the true value of the water content of the soil body and the variable-temperature characteristic index of the test data set; the normalization processing is to normalize the variable-temperature characteristic index and the true value of the water content of the soil body to be a numerical value between [0,1], and the normalization expression is as follows:
wherein: y is ΔT (i) Is the temperature change characteristic index after the i group normalization, y θ (i) The water content value of the soil body after the normalization of the ith group is obtained; delta T i For the i-th group of variable-temperature characteristic indexes in the data set, delta T max For the i-th group maximum variable temperature characteristic index in the data set, delta T min The method comprises the steps of (1) setting the minimum variable-temperature characteristic index of an ith group in a data set; θ i For the water content of the soil body of the ith group in the data set, theta max For the maximum soil moisture content in the data set, theta min The water content of the soil body is the minimum in the data set;
(4) Establishing an artificial neural network model, training, establishing a calibration model suitable for the transient variable temperature fiber bragg grating sensor based on the artificial neural network model, and training a normalized training data set to obtain a trained artificial neural network model;
(5) Inputting the normalized variable-temperature characteristic index of the test data set into a trained artificial neural network model to obtain a soil body water content normalization value predicted by the artificial neural network;
(6) Carrying out return normalization processing on the soil moisture content normalization value predicted by the artificial neural network, and calculating Root Mean Square Error (RMSE) between the return normalized soil moisture content value and the soil moisture content true value; the soil body water content value return normalization treatment is to restore the soil body water content value normalization value to the original order of magnitude, and the expression is as follows:
θ 0 '(i)＝y θ '(i)×(θ max -θ min )+θ min (5)
wherein: θ 0 ' (i) is the return normalization value of the water content of the soil body of the i group; y is θ ' (i) is the water content of the soil body of the ith group obtained based on the artificial neural network modelA normalized value;
(7) Performing artificial neural network model verification, setting expected RMSE, and repeating the steps (4) to (6) if the RMSE in the test step (6) is greater than the expected RMSE; if the RMSE in the test step (6) is smaller than or equal to the expected RMSE, taking the trained artificial neural network model as a calibration model of the measuring point;
(8) The calibration of all measuring points is completed, and a soil moisture content calibration model of the transient variable temperature fiber bragg grating sensor based on the artificial neural network is obtained;
(9) And inputting the measured variable-temperature characteristic index into a soil moisture content calibration model to monitor the moisture content of the soil in situ.
2. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (4), the established artificial neural network model is divided into an input layer, an hidden layer and an output layer, wherein the input layer of the artificial neural network model is set to be a variable-temperature characteristic index of a measuring point and an adjacent measuring point, and the output layer is set to be a true value of the water content of a soil body of the measuring point.
3. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (4), the artificial neural network model is optimized by adjusting the number of unit nodes of the hidden layer and the learning rate.
4. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (4), training of the artificial neural network model is divided into forward propagation and backward propagation, in the forward propagation, an input layer is processed through a unit node of an hidden layer and is transmitted to an output layer, if the output layer cannot obtain expected output, the backward propagation is carried out, and an error signal is returned along the unit node of the original hidden layer.
5. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (4), a transmission rule of the longitudinal temperature inside the transient variable-temperature fiber bragg grating sensor is established by inputting variable-temperature characteristic indexes of adjacent measuring points, and the influence of the longitudinal temperature transmission on the variable-temperature characteristic indexes of the measuring points is removed.
6. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (4), when the artificial neural network model is trained, a mapping relation between the variable-temperature characteristic index and the water content true value is established by using a transfer function, and iterative solution is carried out by adopting a gradient descent method, so that the loss function is minimized, and the artificial neural network model is optimized.
7. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: in the step (7), when the steps (4) to (6) are repeated, the node weight and the threshold between the input layer and the output layer are readjusted when the artificial neural network model is trained.
8. The in-situ monitoring method for the moisture content of the soil body based on the transient temperature-changing fiber bragg grating of the artificial neural network, which is characterized by comprising the following steps of: the accuracy of predicting the water content of the soil body by the artificial neural network model is improved by a data augmentation, model regularization and migration learning method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111610089.XA CN114279973B (en) | 2021-12-27 | 2021-12-27 | In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111610089.XA CN114279973B (en) | 2021-12-27 | 2021-12-27 | In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114279973A CN114279973A (en) | 2022-04-05 |
| CN114279973B true CN114279973B (en) | 2024-03-19 |
Family
ID=80875895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111610089.XA Active CN114279973B (en) | 2021-12-27 | 2021-12-27 | In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114279973B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5316950A (en) * | 1993-01-21 | 1994-05-31 | The United States Of America As Represented By The Secretary Of The Navy | Method for quantitative calibration of in situ optical chemical measurements in soils using soil class and characteristics |
| JP2000121578A (en) * | 1998-10-20 | 2000-04-28 | Syst Sogo Kaihatsu Kk | Apparatus and method for calibration of moisture meter |
| CN102072922A (en) * | 2009-11-25 | 2011-05-25 | 东北林业大学 | Particle swarm optimization neural network model-based method for detecting moisture content of wood |
| CN104034861A (en) * | 2014-06-04 | 2014-09-10 | 中国科学院合肥物质科学研究院 | Method and device for measuring soil bulk density in real time |
| CN108279211A (en) * | 2018-03-25 | 2018-07-13 | 南京大学 | A kind of quasi-distributed system, sensing device and method measuring soil moisture content |
| CN109186811A (en) * | 2018-09-18 | 2019-01-11 | 天津科技大学 | A kind of FBG temperature calibration method based on BP neural network |
| CN110658123A (en) * | 2019-09-23 | 2020-01-07 | 南京大学 | In-situ test method for unsaturated soil permeability coefficient based on optical fiber active temperature change |
| CN112772384A (en) * | 2021-01-28 | 2021-05-11 | 深圳市协润科技有限公司 | Agricultural water irrigation system and method based on convolutional neural network |
| CN112882100A (en) * | 2021-02-25 | 2021-06-01 | 中海石油深海开发有限公司 | Reservoir parameter determination method and device, electronic equipment and storage medium |
| CN113406007A (en) * | 2021-06-16 | 2021-09-17 | 南京大学 | Soil moisture content intelligent monitoring system based on heat pulse identical weak fiber grating array and in-situ calibration method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9228940B2 (en) * | 2012-09-14 | 2016-01-05 | Halliburton Energy Services, Inc. | Systems, methods, and apparatuses for in situ monitoring of cement fluid compositions and setting processes thereof |
-
2021
- 2021-12-27 CN CN202111610089.XA patent/CN114279973B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5316950A (en) * | 1993-01-21 | 1994-05-31 | The United States Of America As Represented By The Secretary Of The Navy | Method for quantitative calibration of in situ optical chemical measurements in soils using soil class and characteristics |
| JP2000121578A (en) * | 1998-10-20 | 2000-04-28 | Syst Sogo Kaihatsu Kk | Apparatus and method for calibration of moisture meter |
| CN102072922A (en) * | 2009-11-25 | 2011-05-25 | 东北林业大学 | Particle swarm optimization neural network model-based method for detecting moisture content of wood |
| CN104034861A (en) * | 2014-06-04 | 2014-09-10 | 中国科学院合肥物质科学研究院 | Method and device for measuring soil bulk density in real time |
| CN108279211A (en) * | 2018-03-25 | 2018-07-13 | 南京大学 | A kind of quasi-distributed system, sensing device and method measuring soil moisture content |
| CN109186811A (en) * | 2018-09-18 | 2019-01-11 | 天津科技大学 | A kind of FBG temperature calibration method based on BP neural network |
| CN110658123A (en) * | 2019-09-23 | 2020-01-07 | 南京大学 | In-situ test method for unsaturated soil permeability coefficient based on optical fiber active temperature change |
| CN112772384A (en) * | 2021-01-28 | 2021-05-11 | 深圳市协润科技有限公司 | Agricultural water irrigation system and method based on convolutional neural network |
| CN112882100A (en) * | 2021-02-25 | 2021-06-01 | 中海石油深海开发有限公司 | Reservoir parameter determination method and device, electronic equipment and storage medium |
| CN113406007A (en) * | 2021-06-16 | 2021-09-17 | 南京大学 | Soil moisture content intelligent monitoring system based on heat pulse identical weak fiber grating array and in-situ calibration method |
Non-Patent Citations (9)
| Title |
|---|
| A dynamic all parameters adaptive BP neural networks model and its application on oil reser voir prediction;YU S W, 等;Applied Mathematics and Computation;第195卷(第1期);66-75 * |
| A soil moisture estimation method using actively heated fiber Bragg grating sensors;Cao, Dingfeng, 等;ENGINEERING GEOLOGY(第8期);142-149 * |
| 一种准分布式内加热刚玉管FBG渗流速率监测方法;段超喆，等;防灾减灾工程学报;第38卷(第03期);504-510 * |
| 不同环境温度下土体含水率主动加热光纤法监测试验研究;郭君仪，等;岩土力学;第41卷(第12期);4137-4144 * |
| 基于BP人工神经网络的土壤含水量预测模型的研究;郭庆春;王素娟;何振芳;;山东农业科学(第12期);17-21 * |
| 基于BP神经网络的土壤水热动态预测模型研究;王宏宇，等;节水灌溉(第07期);11-15+27 * |
| 基于FBG的黄土水分测量技术与应用研究;段新春;中国优秀硕博士学位论文全文数据库（硕士）基础科学辑(第04期);A011-592 * |
| 基于加热光纤分布式温度传感器的土壤含水率测定方法;胡优，等;农业工程学报;第35卷(第10期);42-49 * |
| 青藏高原工程走廊带典型土导热系数实验研究;黄齐;中国优秀硕士学位论文电子期刊工程科技Ⅱ辑(第01期);1-70 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114279973A (en) | 2022-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109186811B (en) | FBG temperature calibration method based on BP neural network | |
| CN113406007B (en) | Soil moisture content intelligent monitoring system based on heat pulse congruent weak optical fiber grating array and in-situ calibration method | |
| Schröder et al. | Wind turbine site-specific load estimation using artificial neural networks calibrated by means of high-fidelity load simulations | |
| CN103994334B (en) | Pipeline Leak flow estimation unit based on KPCA-RBF curve matching and method | |
| CN103808426B (en) | The indirect measurement method of cable core temperature | |
| CN108647807B (en) | River flow prediction method | |
| CN102562047A (en) | System and method for correcting borehole environment under centered array sensing instruments | |
| CN111753250B (en) | One-dimensional unsteady state heat conduction inverse problem method | |
| CN114048790A (en) | Road base layer strain analysis method based on coupling bidirectional LSTM and convolution structure network | |
| CN108491931A (en) | A method of non-destructive testing precision is improved based on machine learning | |
| CN114279973B (en) | In-situ monitoring method for soil moisture content of transient variable-temperature fiber bragg grating based on artificial neural network | |
| CN114912364A (en) | Natural gas well flow prediction method, device, equipment and computer readable medium | |
| CN106228033A (en) | Three-core cable conductor temperature real-time computing technique based on RBF neural | |
| CN106779052A (en) | A kind of distribution cable insulating barrier real-time temperature calculation method based on BP neural network | |
| CN113887142A (en) | Average flow velocity calculation method based on channel cross section | |
| CN111914488B (en) | Data area hydrologic parameter calibration method based on antagonistic neural network | |
| CN106934096B (en) | Method for solving steel core temperature based on surface temperature of overhead conductor | |
| CN107688862A (en) | Insulator equivalent salt density accumulation rate Forecasting Methodology based on BA GRNN | |
| CN117310361A (en) | Power distribution network fault patrol positioning method based on intelligent perception and equipment image | |
| Wang et al. | A piecewise analysis model for electrical conductivity calculation from time domain reflectometry waveforms | |
| CN113945729A (en) | Average flow velocity calculation method based on channel vertical section | |
| CN115468979A (en) | Processing method of soil temperature time sequence | |
| CN115034383A (en) | Network training method, temperature prediction method, device, equipment and storage medium | |
| CN107766972B (en) | Method and device for acquiring energy flow delay time | |
| CN114297940B (en) | Method and device for determining unsteady state reservoir parameters |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |